Journal Logo


Utilizing Computational Oncology to Better Understand AML & MDS Patients

Nalley, Catlin

doi: 10.1097/01.COT.0000553979.53167.65
  • Free
computational oncology; leukemia
computational oncology; leukemia:
computational oncology; leukemia

Relapse is a major challenge when it comes to the treatment of patients with acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS); oncologists are constantly searching for ways to combat this issue while improving their understanding of disease mechanisms.

“We have learned a lot about AML and MDS thanks to advances in DNA-sequencing,” noted Christopher R. Cogle, MD, Professor of Medicine at the University of Florida. “And one of the things we have learned through this effort is that these diseases are much more genetically complex than we previously thought.

“With this new appreciation for the number and variety of genomic abnormalities, we are beginning to see individual AML and MDS cases as unique diseases loosely connected to one another.”

Despite an increased understanding of the genomic makeup of these diseases, response rates, especially in the relapsed/refractory setting, remain poor. “In patients with relapsed or refractory AML or MDS we are seeing response rates of around 20 percent,” Cogle noted. “There is an urgent need for new therapeutics or intelligently repurposed older drugs.”

Recognizing the need for a deeper understanding of the complexities of these diseases, Cogle and his team pursued a new approach. “We sought to utilize computational means to interpret the high number and wide variety of abnormalities that are presenting in our MDS and AML clinic.”

After finding the right software partner and validating the accuracy of the technology through retrospective analysis, the iCare1 prospective clinical study was launched. Findings from this trial were recently presented at the 2018 ASH Annual Meeting (Abstract 3086).

Study Details

Researchers utilized a genomics-informed computational biology modeling (CBM) technique to improve their understanding of the mechanisms of response or relapse after chemotherapy treatment among AML and MDS populations and to hypothesize new treatment approaches.

The investigators recruited 120 patients with AML and MDS to assess the accuracy of CBM prediction through the comparison of computer predictions of treatment response and actual clinical outcomes. Of these patients, 96 had full genomic testing profiles.

Conventional cytogenetics, whole exome sequencing, and array CGH were used to conduct genomic profiling. Disease-specific protein network maps for each patient were created by inputting somatic gene mutations into the CBM program, Cogle explained.

A digital library of FDA-approved drugs was generated for the technology by “programming each agent's mechanism of action determined from published literature,” researchers outlined. “Digital drug simulations of the patient's choice of therapy were tested at varying doses and predicted efficacy of the drugs were measured as a function of a disease inhibition score, defined as the degree to which disease pathways and phenotypes (cell proliferation and viability) were mathematically returned to a mutation-free state.”

Based on length of follow-up and minimum treatment exposure threshold, 50 patients were eligible for evaluation. Among these patients, 61 treatments were administered. Researchers reported that the CBM maps of relapsed samples from these patients “accurately matched the patient's nonresponse of treatment at relapse in 90 percent of patients and identified mechanisms for chemoresistance.”

“By applying this computational biology method and digital drug simulation we found several things,” Cogle noted. “Chief amongst them is the capability of this computational method to be employed in the clinic and provide highly accurate predictions of treatment response in our AML and MDS patients.”

Implications, Next Steps

Using computers in oncology is an emerging area not only in research, but also for clinical application, noted Cogle. However, he emphasized the importance of comprehensive testing of these methods before marketing to patients with cancer.

“The iCare1 study and the work preceding it represents, firsthand, the kind of stepwise, rigorous clinical trial testing that should be done for predictive methods,” Cogle elaborated. “We are not only demonstrating a feasible and accurate computational method, but we are also highlighting a path towards responsible technology development.”

While the next step for the computational model tested in iCare1 is to assess it in a randomized clinical trial, Cogle and his team are also focused on educating physicians and patients about this type of technology so they are “empowered to appreciate the results that come out of computational predictions.

“Computational oncology is called a ‘black box’ only by those who don't know how to open the box and read what's inside,” Cogle said. “My group is committed to assisting physicians and patients to better understand the methods being used so that they can determine whether or not it is an appropriate application for their case.”

Looking forward, Cogle stressed the important role computational oncology will play in the field of oncology. “Interpreting one gene mutation can be challenging and time consuming, interpreting multiple gene mutations is sometimes impossible in our busy practice setting,” he noted. “We have to embrace computational oncology because of limitations not only on our time, but also on our human cognitive capacity to parallel process the tangle of abnormalities that we find in cells of our cancer patients.

“Results like the ones coming out of the iCare1 study are encouraging and demonstrate that we now have access to technology that is not only feasible, but also accurately predicts treatment response.”

Catlin Nalley is associate editor.

Wolters Kluwer Health, Inc. All rights reserved.
    Home  Clinical Resource Center
    Current Issue       Search OT
    Archives Get OT Enews
    Blogs Email us!